• Introduces parallel batch-dynamic algorithms for maintaining low out-degree orientations in undirected graphs. • Achieves polylogarithmic depth with high probability, focusing on minimizing work per edge. • First algorithm achieves asymptotically optimal orientation with optimal expected work, improving prior work by log factor. • Second algorithm offers O(c log n) orientation with expected worst-case O(√log n) work per edge, matching best sequential worst-case. • Third algorithm provides O(c + log n) orientation with O(log^2 n) expected worst-case work, surpassing Ghaffari & Koo’s O(log^9 n). • Results advance parallel dynamic graph algorithms, enabling efficient updates for graphs with bounded arboricity.

Article Summaries:

  • Researchers have introduced a suite of faster parallel batch‑dynamic algorithms for maintaining low‑out‑degree orientations in undirected graphs. All methods achieve polylogarithmic depth with high probability while reducing per‑edge work. The first algorithm delivers an asymptotically optimal orientation with expected amortized work that improves Liu et al.’s (SPAA 22) bounds by a logarithmic factor. The second offers an (O(c\log n)) orientation (where (c) bounds arboricity) with expected worst‑case (O(\sqrt{\log n})) work per edge, matching the best sequential worst‑case result in expectation. The third achieves an (O(c+\log n)) orientation with (O(\log^2 n)) expected worst‑case work, vastly improving on Ghaffari and Koo’s (SPAA 25) (O(\log^9 n)) bound. These advances advance parallel graph‑orientation efficiency in distributed and cluster computing.

Sources: